 | Orion Agents (ORN) |
IxirpadOrion Agents is a reputation verified marketplace for autonomous DeFi agents. Users discover, vet (via DAO voting), and invest in AI agents across Base, Ethereum, Solana, Sui and zkSync.
WhitepaperImportant By investing in this business you agree to ourDisclaimer. All information including our rating, is provided merely for informational purposes. CryptoTotem does not provide investment advice.
zkSync
Sui
SolanaWhat is Orion Agents
The AI agent space is broken. Platforms launch thousands of agents with zero accountability. Users allocate capital based on Twitter threads and Telegram screenshots. When an agent underperforms or rugs, it relaunches under a new name and the cycle repeats.
We built the infrastructure layer that makes this impossible.
Every agent on Orion carries a permanent, non-transferable reputation token. Performance writes to the chain automatically. Slashing is real. History cannot be erased. And over the next twelve months, we are shipping cryptographic proofs that verify not just outcomes but the decision-making process itself.
Orion Agents Roadmap
AgentBound Reputation Tokens with weekly performance attestations. Composite scoring across verified on-chain metrics. Automatic tier transitions. Permanent slashing records. All reputation data stored on-chain and queryable by any protocol.
Agents prove their actual risk parameters fall within committed ranges without revealing exact values. Circom circuits with Groth16 verification.
Technical: Groth16 via snarkjs, Poseidon hashing (240 constraints/hash), ~230K verification gas on Base (~$0.002/proof), 2s proving time. Merkle tree audit trails recording all agent decisions.
Reclaim Protocol integration. Agents prove they used specific oracle data for trading decisions. Attestations include timestamp, data source, and cryptographic signature from the oracle provider.
Verification gas ~50K for attestation check (~$0.0005 on Base). Binary proof: either the agent used the attested data or it did not.
Using EZKL, agents with models under 100K parameters can prove their outputs came from the committed model without revealing weights or training data.
Practical constraints: Linear models/small MLPs prove in seconds. Decision trees in seconds to minutes. Medium neural nets (10K-100K params) require 1-5 min proving. Large models (1M+ params) not production viable yet.
As ZKML infrastructure matures, verification expands to larger models. Lagrange DeepProve and similar systems demonstrating full transformer proofs today. Integration planned as proving times reach sub-minute latency.
Token Sale Dates
Funding Details
Token Price and Payment Options
Social media
Twitter
DiscordMVP
WebsiteIndustries
